blob: b979b7573ae05f92d8045313fc10020542a43446 (
plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
|
# frozen_string_literal: true
module Banzai
module Filter
class MarkdownPostEscapeFilter < HTML::Pipeline::Filter
LITERAL_KEYWORD = MarkdownPreEscapeFilter::LITERAL_KEYWORD
LITERAL_REGEX = %r{#{LITERAL_KEYWORD}-(.*?)-#{LITERAL_KEYWORD}}.freeze
NOT_LITERAL_REGEX = %r{#{LITERAL_KEYWORD}-((%5C|\\).+?)-#{LITERAL_KEYWORD}}.freeze
SPAN_REGEX = %r{<span>(.*?)</span>}.freeze
CSS_A = 'a'
XPATH_A = Gitlab::Utils::Nokogiri.css_to_xpath(CSS_A).freeze
def call
return doc unless result[:escaped_literals]
# For any literals that actually didn't get escape processed
# (for example in code blocks), remove the special sequence.
html.gsub!(NOT_LITERAL_REGEX, '\1')
# Replace any left over literal sequences with `span` so that our
# reference processing is short-circuited
html.gsub!(LITERAL_REGEX, '<span>\1</span>')
# Since literals are converted in links, we need to remove any surrounding `span`.
# Note: this could have been done in the renderer,
# Banzai::Renderer::CommonMark::HTML. However, we eventually want to use
# the built-in compiled renderer, rather than the ruby version, for speed.
# So let's do this work here.
doc.xpath(XPATH_A).each do |node|
node.attributes['href'].value = node.attributes['href'].value.gsub(SPAN_REGEX, '\1') if node.attributes['href']
node.attributes['title'].value = node.attributes['title'].value.gsub(SPAN_REGEX, '\1') if node.attributes['title']
end
doc.xpath(lang_tag).each do |node|
node.attributes['lang'].value = node.attributes['lang'].value.gsub(SPAN_REGEX, '\1') if node.attributes['lang']
end
doc
end
private
def lang_tag
if Feature.enabled?(:use_cmark_renderer, default_enabled: :yaml)
Gitlab::Utils::Nokogiri.css_to_xpath('pre')
else
Gitlab::Utils::Nokogiri.css_to_xpath('code')
end
end
end
end
end
|
